Just like every company needed a website in 2001, and businesses spent big bucks on making their own mobile apps in 2010, the next decade saw marketing budget spent on the software and data scientists to collect and dissect every data-point possible – from search and email to social media interactions and ad impressions.

But then, what to do with it? The dream scenario for businesses is to discover miraculous hidden patterns in their marketing data which can be magically rearranged into multi-touch attribution models, closing the gap between customer behaviour and business bottom-line.

That said, many marketers are fighting the good fight, but understandably lost in the detail of the ‘what’ and the ‘how’, without clear oversight on the bigger picture, and this is crucial to making the most of your data:

● What are the goals of your data collection and analysis?
● What are the performance indicators of your goals, and the metrics you need to measure to understand your progress?
● What will you do with those insights when you have them?

Let’s explore a four-level effort vs. value matrix which I use with clients to help understand where they sit in terms of data-driven marketing maturity.

This is a maturity model popularised by Gartner, and which many data analytics organisations use as a basis for jumping straight into the deep-end with AI and machine learning. What I believe we miss is some focus on the initial phases of maturity: how do you get started, first creating a scalable baseline for predictive intelligence based on your goals?

Level 1: Descriptive data – ‘this has happened’

Companies at this stage measure the quantity of email opens, the number of visitors to their website, and the number of likes they get on Facebook.

This describes your customer’s movements, and it’s a fundamental baseline for getting a deeper understanding of customer behaviour.

Level 2: Diagnostic data – ‘this has happened because of [x]’

If you’re a marketer on this level, you can understand the relationship between an event and an outcome. For example, you see that this post caused that amount of likes, or this amount of email opens, with that content, on this day.

Diagnostic data gives you a greater understanding of what you need to do with descriptive data.

Level 3: Predictive data – ‘this is what will happen’

Then, you can begin to pre-empt marketing outcomes. If that type of content generated this amount of likes or opens, then if you did something similar again, you can accurately predict that it will have a similar result.

Level 4: Prescriptive data – ‘this is how we’re going to make it happen.’

And here we have the holy grail; an intelligent, self-optimising model which takes in new data to re-predict and re-prescribe actions based on learning and attributing value.

Of course, almost all brands strive for level 4, but it’s no easy journey.

The main challenges they see exist in fragmented systems and an inability to easily identify patterns between datasets and teams. Investing in data visualisation and data management solutions is a valuable step, but one caveat I see when speaking to companies about their data-driven marketing strategy is that adopting highly capable technology comes too far ahead of a plan on what to do with it – when the two should really run in parallel.

To avoid false starts, marketers can validate their goals and create the rules and the guardrails for their data-driven journeys. And this is when dashboarding and data management solutions can be hugely valuable.

What actionable analysis looks like in practice

Companies I work with at Emark start off by giving goals to their collection and application of customer data: so instead of a strategy for everything-at-once, that means making plans to turn their abandoned carts into conversions, for example.

In directing their focus towards analysing web and email customer data, the eCommerce team notices something strange happening on their webshop; a considerably higher number of checkout pages visited than ‘add-to-basket’ actions taken for that month.

To discover why this was happening, and to make sure customers weren’t somehow checking out without buying, they looked for a golden intersection between their web data, and their email insights, using Datorama’s rich cross-channel dashboarding capabilities to connect the dots.

In doing this, they discovered a pattern: it turned out that in the same week, an abandoned cart reminder was sent out – and a backlog of carts from an earlier period was converted.

From smart descriptive data to the prescriptive data dream

Without this level of focus for using a platform like Datorama, the team may never have diagnosed this disconnect between webpage visits and conversions. They would have had to rely on best-guesses or manual uploads between siloed email and web departments, losing those key insights needed to help them get to the next level of data-driven marketing.

So what does the future look like for brands who get the basics right with Datorama? In the next couple of years, the marketing team may have moved onto automating and predicting next-best-actions. They can do this by weighing in on the products someone sees on the website and calculating their product affinity based on their behaviour – similar items viewed, return visits, previous purchases.

They can add context to their calculations, using factors such as previous engagements, clicks and buying behaviour to predict whether a customer will buy this product right now or requires inspiration first. They can optimise not only their next-best-actions but their preferred customer channel-mix too.

How can ambitious commerce brands get there, in summary? It’s all about taking a phased approach to scaling the mountain of customer data:

Have this baseline in place for your analytics, and you can achieve a whole new level of marketing success long before you reach the predictive data summit.

Ready for more inspiration about how to put your data to work across your business? Read Sander’s event takeaways from Salesforce World Tour.